For the preparation of suitable systems, acetone -dried powders from rabbit muscle and washed red blood cells of the horse were extracted twice with ten times their weight of 0 -05 M 'tris' buffer (tris(hydroxymethyl)-aminomethane) of pH 7. The extracts were heated at 55-60 C. for 5 min., freed from the resulting coagulum, and the fraction precipitated between 0-3-0-5 saturation with ammonium sulphate was suspended in water and dialysed. After rejection of the insoluble globulins, the freeze-dried material contains most of the pentosephosphate isomerase of the original extract, and remains active for months when stored at 20. For the preparation of the barium salt of the ketopentose phosphate, the muscle preparation (30 mgm.) was incubated at 37 with a solution of 1 gm. (2,500 pi-moles) barium ribose-5-phosphate in 25 ml. water at pfL 7. The ketopentose formed was measured by the Dische reaction (method of Axelrod and Jang3), until no further increase occurred (90 min.) and it remained unchanged at 930 pi-moles (37 per cent) for a further 30 min. After deproteinizing with perchloric acid, careful neutralization, and removal of aldoses with bromine and barium carbonate, the phosphate esters were adsorbed on a 'Dowex-1 (formate)' column and eluted with 0-1 M formic acid in 0-1 M sodium formate. By careful neutralization with barium hydroxide and precipitation with 4 vol. of ethanol, the vacuum-dried barium salt (357 mgm.) was obtained with a purity of 93 per cent based on ketopentose and 85 per cent based on ester-phosphorus content (800 yt. moles ketopentose phosphate, or 32 per cent of the total initial pentose). At least 90 per cent of the total ketopentose was alkali-labile, and the rate of hydrolysis in N acid at 100 (k = 5-2 x 10~3) also indicated a 5-position for the phosphate. The ratio of intensities in the orcinol reaction at 540/670 mjj. we found to be 0-90-1-00 for chromatographically purified synthetic ribulose and 0-45-0-50 for xylulose. The free ketopentoses liberated by purified acid phosphatase from the above barium salt gave an intermediate value (0-55-0-60). The dephosphorylated de-ionized sugars separated sharply with 88 per cent recovery on a 'Dowex-1 (borate)' column7 into two peaks, of which the first (51 per cent of total ketopentose in eluate) corresponded exactly in colour reactions and position on elution with pure xylulose and the second (49 per cent) with pure ribulose. No interconversion of the pure sugars similarly processed was detectable.The enzyme prepared from red cells gave a similar proportion (43 per cent) of total ketopentose to pentose at 'equilibrium'. But the barium salt, prepared and treated as above, had orcinol ratio 0-78, and separation on the column showed that it was substantially the ribulose ester (peak I = 10 per cent, peak II = 90 per cent of total ketopentose ; orcinol ratios 0-50 and 0-90 respectively). This barium salt, when incubated with the muscle preparation, lost 59 per cent of its total ketopentose (by conversion to ribose phosphate), while the remaining ketopentose was largely the xylulose ester (peak I, 71 per cent; peak II, 29 per cent of total bromine-resistant pentoses ; orcinol ratios 0-50 and 0-98 respectively). At least one enzyme in addition to the pentose phosphate isomerase is therefore active in the muscle preparation, and converts ribulose-5-phosphate to the xylulose ester. Since no 3-ketopentose6 formation was detected, the enzyme is presumably a type of 'waldenase', causing inversion at C(3) of the pentose, and resembling that recently suggested as active in Pseudomonas hydrophila8.